Forecasting with high uncertainty and long-time horizons still challenges researchers and practitioners. A widely adopted method in knowledge sharing and forecasting based on experts is the Delphi method and its offspring, the Real-Time Delphi. While the traditional Delphi method already is intensely investigated, the Real-Time Delphi is still evolving, and no dominant design has been found yet. A problem arising in both variants of the Delphi method, are high drop-out rates between rounds. This paper applies a design science research approach to motivate the need for social design elements from literature and derives design principles for Real-Time Delphi platform. Based on the design, we implement and evaluate a prototype in an online experiment as well as an IT artifact in a field study. We find significant supporting evidence, that (the promise of) positive social reputation increases commitment, and therefore subsequent platform engagement of our Real-Time Delphi survey. Our findings, therefore, contribute valuable design knowledge for Real-Time Delphi platforms. Moreover, we provide advice on how to raise retention in knowledge sharing systems.